Treatment-resistant depression (TRD) is a major public health problem;more than 50% of depressed patients fail to remit after one adequate treatment with an antidepressant, and approximately 35% remain symptomatic after receiving two antidepressants. Novel pharmacologic (e.g., atypical antipsychotic augmentation) and nonpharmacologic (e.g., vagus nerve stimulation) interventions have been developed, but there is little information on the long-term comparative effectiveness and safety of existing treatment options in the larger and more diverse groups of TRD patients who seek care in routine clinical settings. Randomized controlled trials may not fill this knowledge gap in a timely fashion because of their relatively small sample size, selective study population, short-term follow-up, high cost, time consumption, and limited treatment strategies considered. On the other hand, electronic healthcare databases record clinical encounters of a large number of patients, making it possible to study real world utilization patterns, and comparative risks and benefits of various therapeutic options for TRD. Because these databases accurately chronicle changes in antidepressive therapies, which are strong indicators for TRD, studies that use these databases have a great potential to complement randomized trials to provide the much needed clinical information to help improve quality of care for TRD patients. We propose a cohort study to use the administrative claims data from a large non-profit health plan with longitudinal follow-up and linkage to full-text medical records to evaluate the feasibility of identifying TRD patients in claims databases. To understand the state of clinical practice, we will also examine the treatment sequences of antidepressive therapies and utilization patterns of existing treatment strategies for TRD. This study will greatly enhance our capacity to use electronic healthcare databases to assess a wide range of critical issues surrounding therapeutic options for TRD. The project will use data from one of the health plans participating in the HMO Research Network, a 15-year old consortium of 15 U.S. health plans that serve 11 million geographically and demographically diverse members. Because all databases in the Network share identical data specifications, algorithms developed in this study can be directly applied to the entire Network, laying the groundwork to conduct subsequent studies with these databases for timely and state-of-the-art assessment of the long-term comparative effectiveness and safety of various treatment strategies for TRD. By identifying and targeting patients with TRD, the study may also have direct clinical implication with regard to delivery of care, because the research team, the health plan and the delivery system involved in the study have long standing and close relationships to integrate research findings into policy and clinical practice to improve quality of antidepressive treatment of their members and patients. PUBLIC HEALTH RELEVANCE: We will develop and validate algorithms to identify treatment-resistant depression in administrative claims databases. The study will lay the groundwork to conduct subsequent studies with these databases for timely and state-of-the-art assessment of the long-term comparative effectiveness and safety of various therapeutic options for treatment-resistant depression, and eventually lead to better quality of care.